Habitat selection by Black kite breeders and floaters: Implications for conservation management of raptor floaters ⇑ Alessandro Tanferna, Lidia López-Jiménez, Julio Blas, Fernando Hiraldo, Fabrizio Sergio
Department of Conservation Biology, Estación Biológica de Doñana, Consejo Superior de Investigación Científicas, Sevilla 41092, Spain
abstract
Preserving large predators is important but challenging because these species are typically wide-ranging, select multiple habitats at different scales and often present spatial or habitat separation between the breeder and floater sectors of a population. In addition, most of our knowledge on raptor floaters’ habitat requirements comes from large solitary species, whose floaters often occupy temporary settlement areas spatially separate from breeding locations. Here, we examine space and habitat use by a loosely colonial, wetland-dependent raptor, the Black kiteMilvus ( migrans), in a population where floaters co-exist with Keywords: territory holders, enabling a direct comparison of their habitat preferences. The study was conducted Habitat selection in Doñana National Park (South-Western Spain), a seasonally drying marshland currently surrounded Black kite Milvus migrans by intensive agriculture and rice-fields. Intensive radio-tracking revealed that breeders and floaters Radio-tracking selected and avoided the same habitats despite a radical, four-to-eight fold difference in their home- Habitat restoration range dimensions: all kites over-selected open habitats suitable for their aerial foraging modes and Non-breeders avoided woodland and farmland. These results suggest a continuum of raptor population structures rang- ing from solitary species whose floaters select different habitats than breeders and are concentrated in spatially separate settlement areas, to colonial and semi-social species whose floaters fully coexist with breeders with shared habitat preferences. Both extremes of this continuum will pose challenges for con- servation management. In solitary species, special conservation efforts may be required to identify and manage temporary settlement areas, while in gregarious species, the larger ranges of floaters may expose them to different threats than breeders, whose occurrence and consequences may be subtle to identify. .
1. Introduction (e.g. David Smith, 1993; Ferrer and Harte, 1997; Crabtree and Shel- don, 1999; Balbontín, 2005). The latter adds complexity to strate- Preserving and managing large vertebrate predators is becom-gic management targeting long-term population persistence, ing increasingly important as a way to maintain high levels especialof bio-ly because non-breeding animals are difficult to study diversity (Estes et al., 2011), but poses special challengesdue for to their cryptic behaviour, differential habitat selection, spatial several reasons. First, these species are characterised by largeseparation from breeders, or potential long-distance dispersal (e.g. home-ranges, which cannot be easily encompassed within pro-Zack and Stutchbury, 1992; Rohner, 1997; Whitfield et al., 2009a; tected areas (e.g. Newton, 1979; Clark et al., 1999; Ray Penterianiet al., et al., 2011). As a result, there is little knowledge on the 2005). Second, they frequently select habitat features at multipledifferences in habitat choices between the breeders and floaters of scales, from the micro-scale to the landscape-level (Sánchez-Zapaa tapopulation, and conservation planning is often biased to protect and Calvo, 1999; Thompson and McGarigal, 2002; Ciarniello etthe al., habitats preferred by the breeding sector of predator popula- 2007), which requires broad-level management plans (e.g. tions (e.g. Real and Mañosa, 1996; Whitfield et al., 2006a). Also, Whitfield et al., 2006a). Third, they may use different habitatsmost atof the (scarce) available knowledge is heavily biased towards different times of the year (Boal et al., 2005; Schmitz largeet al.,species of solitary birds of prey, whose floaters are typically 2010). Finally, the populations of large predatory vertebrates concentare rated in so-called ‘‘temporary settlement areas’’, where frequently composed of a sector of territorial breeders, often theycon- select different habitats than breeders (Ferrer and Harte, centrated in resource-rich sites, and a sector of non-breeding 1997;indi- Balbontín, 2005; Caro et al., 2011; Penteriani et al., 2011). viduals, frequently located far away from the breeding groundsAs a result, little is known of smaller species with different social systems, such as colonial or loosely colonial species. ⇑ Corresponding author. Tel.: +34 605 482244. Because of all the above, there is a high need for further multi- E-mail addresses:[email protected], [email protected] (A. Tanfer- scale habitat selection studies on both breeding and non-breeding na). individuals of predatory vertebrates, particularly of gregarious inantor conspecifics (Sergio et al., 2009; Blas and Hiraldo, 2010; Blas semi-gregarious species. Here we provide such a study by examin-et al., 2011). Sexual role division during reproduction follows the ing the space and habitat requirements of the breeders and floatersusual scheme for raptors (Newton, 1979): the male provides most of a semi-social raptor, the Black Milvuskite (migrans). In particu- of the prey for the female and offspring while the female performs lar, we use data from a 3-year radio-tracking study on themost Black of the incubation, brooding and nest guarding. kite population of Doñana National Park (South-Western Spain), one of the most renowned and biodiversity-rich reserves of Europe.2.2. Study area Our study system is a good model for the goals presented above due to several reasons. (1) The Black kite is a semi-gregarious Therap- study was conducted in Doñana National Park, located tor, which in Doñana mainly breeds in loose colonies (Sergiowithin et al., the estuary of the Guadalquivir river, along the coast of 2005). (2) In this population, floaters coexist with breeders the(BlasAtlantic Ocean in South-Western Spain (6°120–6°400W, et al., 2009; Sergio et al., 2009, 2011a). This allows the study36°48 0–37of °200N). The five main macro-habitats observed in the habitat selection by the two status categories while controllingpark include: (1) seasonally drying marshland (hereafter ‘‘marsh- for differences in habitat availability, avoiding the problem of land’’),com- (2) Mediterranean scrubland or grassland with scattered paring the decisions by groups of individuals occupying separatecork oaks Quercus( suber) (hereafter ‘‘dehesa’’), (3) extensive scrub- areas characterised by different landscapes. (3) Doñana Nationalland on sandy soil (hereafter ’’scrubland’’), a mixture of different Park is an island of semi-natural vegetation subject to dynamicdegradation stages of autochthonous Mediterranean scrubland habitat management and transformation. Outside the park, drain-(Castroviejo, 1993), including patches dominatedPistacia by lentis- age of the seasonal marshes in the second half of the 20thcus centuryand Myrtus communisor byHalimium halimifolium, Ulex spp., has generated a matrix of intensive farmland, dominated byStauraca rice nthus genistoidesand Erica spp.; (4) mobile sand dunes fields to the north-east, whose suitability for wildlife species alongis lar- the ocean coast, and (5) extensive forests of stonePinus pine gely unknown. Inside the protected area, all habitats are tradition-pinea and smaller woodlots dominated by Cork oaksEucalyptus or ally actively managed (e.g. Fernández-Delgado, 2005). For spp. trees (Castroviejo, 1993). A mosaic of intensively cultivated example, large patches of forest have been recently thinned landsor re- and rice fields surrounds the park. moved, while the hydrology of the seasonal marshes that charac- terise the park is subjected to a recently implemented large-scale2.3. Field methods restoration program (Project ‘‘Doñana 2005’’: García Novo and Marín Cabrera, 2005a), which included the restoration of variousBetween 2007 and 2009 we trapped 38 Black kites by cannon- sites totalling more than 502 ofkm seasonal marshland which netting (Fig. 1) and equipped them with a conventional backpack had been originally converted to agriculture (García Novo transmitterand (TW-3 of 15 g; life expectancy = 1.4 years; Biotrack Marín Cabrera, 2005b; Santamaría et al., 2005; Martín-LópezLtd., Wareham Dorset, UK), which was fitted with a Teflon harness et al., 2011). The above described habitat changes and active( Kenward,man- 2001). The sex, status and sampling period of tracked agement inside and outside the park call for more solid knowledgekites are specified in Table 1. Kites were monitored every 3–4 days of the habitat preferences of key species such as Black kites,and which all locations, obtained by triangulation, were GIS mapped are the most abundant large predators in the park and whichthrough de- the software ArcView 3.2 (ArcView GIS, Redlands, CA, pend heavily on woodland for nesting and marshland for huntingUSA). In each tracking day, all marked kites were searched simul- (Sergio et al., 2011b). Understanding habitat preferences of taneouslykey while driving along a network of paved and dirt roads indicator species could be fundamental to forecast future impactscovering the entire park and its surroundings, thus sampling areas of habitat management and to implement more efficient post-both close and far from nest concentrations. Also, the starting point intervention monitoring. and sequence of survey roads were varied each time, in order to Given all the above, here we: (1) examine the home rangeavoid and biasing the tracking data towards certain areas (e.g. towards habitat selection of kites of different sex and status (breedersnest vs.concentrations). Using an area accumulation curve, we found floaters) and (2) propose potential management guidelines basedthat locations sampling saturation was reached for an average on the obtained results. threshold of 40 fixes per individual and all individuals were radio-located more than 40 times. Radio-tracking and the intensive demographic monitoring of 2. Methods the breeding and non-breeding sectors of the population (Sergio et al., 2009, 2011a) allowed us to determine the breeding status 2.1. Study species of all radio-tagged individuals. These included 12 breeding males, 12 breeding females and 14 floaters (eight males and six females). The Black kite is a medium-sized, monogamous, migratory Breedersrap- were defined as individuals holding a territory with a tor. It is an opportunistic, aerial predator typical of open habitatspartner and building a nest. All trapped birds were sexed by molec- (Viñuela and Sunyer, 1992; Blanco and Viñuela, 2004), adeptular atanalysis of a blood sample (Ellegren, 1996). exploiting temporary situations of overabundance of relatively easy prey (Hiraldo et al., 1990). In our study population, all 2.4.individ- GIS and statistical analysis uals are migratory and remain in Doñana from March to August, where they mostly breed as monogamous pairs (Sergio et Foral., each kite, we estimated the home range size and configura- 2007). The local breeding density can be very high (fromtion 1 throughto the following three indices: (1) the Minimum Convex 30 pairs/km2, Sergio et al., 2005, 2011b; authors’ unpublished Polygonre- (MCP), (2) the Kernel Density Estimator (KDE) at 95%, 75% sults) and most pairs could be considered to nest within anda very50% contours, calculated with a least-squares cross-validation large, loose colony. Diet composition is very heterogeneous (LSCV)and procedure and a smoothing factor (Seaman and Powell, dominated by wetland birds and their nestlings, crayfish, rabbits1996) and (3) the mean distance of all fixes from the home-range (Oryctolagus cuniculus) and carrion (Hiraldo et al., 1990; Viñuelacentroid (hereafter ‘‘distance to centroid’’), calculated through the and Veiga, 1997). Floaters are generally young birds (1–7 Animalyears Movement extension for ArcView (Hooge and Eichenlaub, old, Blas et al., 2009) physiologically capable of reproducing200 but0). However, to avoid reporting redundant results, for the anal- apparently displaced from the breeding sites by older, more ysesdom- of habitat selection we only show the models based on the Fig. 1.Main trapping sites (circles) and areas of concentration of Black kite nests (in grey) around the seasonal flooded marshland of Doñana National Park (South-Western Spain) and its surroundings. The areas highlighted in grey include more than 90% of the nests used by Black kites in any given year. The other pairs breed in single, isolated nests or in loose colonies of 2–3 nests scattered around the rest of the landscape. The continuous line represents the border of the cumulated National and Natural Park (marine portion excluded), whose location in Europe is portrayed in the inset.
Minimum Convex Polygon. Models based on Kernel estimatorswhich combination of habitat variables discriminated between gave the same results (results of analysis not shown). the 38 kite home ranges and 38 randomly-plotted home ranges Habitat composition was evaluated by accessing a 1:10 of000 the same shape. The latter were generated by the following land-use map provided by the LAST-EBD group (http://last-ethree-stbd. ep procedure: (1) we plotted the centroid of all 38 kites blogspot.com/). Based on such map, land-uses were initiallyhome ranges; (2) an equal number of random points was generated classified according to the categories listed in Table 2. However,through the Animal Movement extension of the GIS software to reduce the high frequency of zero values obtained for (severalHooge and Eichenlaub, 2000); (3) each kite home range was infrequently used habitat types (Aebischer et al., 1993), we pooledshifted so that its centroid would now coincide with one of the ran- the initial habitats into five coarser-level macro-habitats judgeddomly as plotted centroids. This generated 38 randomly plotted home potentially important for kites on the basis of accumulated knowl-ranges that maintained the same shape of the originally observed edge on the population nesting and foraging behaviour: seasonalranges, allowing us to study habitat selection while controlling marshland, scrubland, dehesa, intensive farmland, and woodland.for home range size. To gain an understanding of kite ranging behaviour and habitatFinally, to investigate finer-scale habitat selection, we: (1) plot- selection, we focused our analyses on three aspects: (1) the tedspatial a 100-m buffer around each kite locationn = 1980 ( locations) extent and configuration of the home range (hereafter ‘‘homeand an equal number of randomly generated locations; (2) calcu- range analysis’’); (2) the habitat composition of a whole latedhome the percentage habitat extent in each buffer using the Arc- range and its comparison with local availability (hereafter View‘‘lar- extension Patch Analyst (Elkie et al., 1999); and (3) ger-scale habitat selection’’); (3) the habitat composition aroundcompared the habitat composition around kites and random loca- each individual radio-location and its comparison with local avail-tions by means of linear mixed models (LMM) where individual ability (hereafter ‘‘finer-scale habitat selection’’). The first analysiskite identity was fitted as a random factor (Zuur et al., 2009). The focuses on the spatial requirements of different types of individu-measure of 100 m for the buffer was arbitrarily chosen because als (males vs. females, breeders vs. floaters), the second on thekites hab- are aerial hunters that patrol large areas while typically soar- itat-based selection of whole home ranges, and the third oning theand gliding at a minimum altitude of 20–30 m, thus scanning a finer-scale selection of habitats within a home-range. landscape portion rather than a single point. For the home range analysis, we used one way ANOVAIn all models, we added Year as a covariate to control for annual (Lehmann and Romano, 2005) to compare the home range variatiosize ns in ranging behaviour. Although we are aware that raptor and distance to centroid among breeding males, breeding femaleshome ranges can vary seasonally (e.g. Newton, 1979; Bosch et al., and floaters. For larger-scale habitat selection, we used a logistic2010), sample size limitations precluded the possibility to examine regression (GLM with binomial errors; Zuur et al., 2009) tseasonalo test (within-year) variations. To reduce such confounding Table 1 Sex, status (breeder vs. floater) and sampling period of 38 adult Black kites radio-tracked in Doñana National Park (South-Western Spain). All individuals were still alive when the radio-signal was lost through battery exhaustion.
Individuals Year monitored Sex Status Tracking period Number of locations M1-B 2007 Male Breeder 2 May–29 July 47 M2-B 2007 Male Breeder 11 May–23 July 69 M3-B 2007 Male Breeder 19 April–26 July 50 M4-B 2007 Male Breeder 20 April–16 July 48 F1-B 2007 Female Breeder 24 April–10 July 57 F2-B 2007 Female Breeder 13 May–19 July 53 F3-B 2007 Female Breeder 25 April–27 July 67 M5-F 2007 Male Floater 02 May–23 July 73 M6-F 2007 Male Floater 21 April–26 July 55 M7-F 2007 Male Floater 30 April–29 July 53 F4-F 2007 Female Floater 29 April–23 July 60 F5-F 2007 Female Floater 30 April–27 July 46 F6-F 2007 Female Floater 30 April–26 July 51 F7-F 2007 Female Floater 24 April–21 July 73 M8-B 2008 Male Breeder 17 April–25 July 58 M9-B 2008 Male Breeder 13 March–25 July 51 M10-B 2008 Male Breeder 28 March–25 July 37 M11-B 2008 Male Breeder 15 March–27 June 45 M12-B 2008 Male Breeder 15 June–26 July 40 F8-B 2008 Female Breeder 26 March–27 June 80 F9-B 2008 Female Breeder 17 April–27 July 59 F10-B 2008 Female Breeder 26 March–27 June 57 F11-B 2008 Female Breeder 26 March–02 June 45 F12-B 2008 Female Breeder 17 April–28 July 57 F13-B 2008 Female Breeder 24 April–28 July 63 F14-B 2008 Female Breeder 24 April–28 July 60 M13-F 2008 Male Floater 13 April–28 July 53 M14-F 2008 Male Floater 15 April–27 July 43 M15-F 2008 Male Floater 28 March–23 July 38 M16-B 2009 Male Breeder 18 March–22 July 52 M17-B 2009 Male Breeder 21 April–15 July 44 M18-B 2009 Male Breeder 20 March–21 July 55 F15-B 2009 Female Breeder 7 April–15 July 45 F16-B 2009 Female Breeder 21 March–05 June 41 M19-F 2009 Male Floater 21 March–23 July 43 M20-F 2009 Male Floater 7 April–21 July 47 F17-F 2009 Female Floater 8 April–06 July 41 F18-F 2009 Female Floater 6 April–21 July 66
Table 2 et al., 2009). In this method, pairs of strongly intercorrelated vari- Environmental variables measured for the home ranges of 38 Black kites andables 38 r(> 0.6) were considered as estimates of a single underlying randomly generated home ranges. Variables were later pooled into a smaller number factor. Only one of the two is retained for analysis, usually the one of descriptors used for analysis (see Methods). likely to be perceived as more important by the study organism. Of Variable Description the remaining variables, only those for which significant univariate % Water % Extent of water bodies excluding the seasonal marshlanddifferences p( < 0.05) were detected between real and random % Rice pounds % Extent of rice fields locations were included in multivariate analyses. When building % Farmland % Extent of intensively managed farmland GLMs and LMMs, all explanatory variables were fitted to a (maxi- % Dumps % Area occupied by rubbish dumps % Dunes % Extent of mobile sand dunes mal) model, extracted one at a time from such maximal model % Marshland % Seasonally drying marshland and the associated change in model deviance assessed by a likeli- % Dehesa % Grassland or scrubland with scattered oak trees hood ratio test (Zuur et al., 2009). At each step, we also calculated % Scrubland % Mediterranean scrubland (matorral) the AICc (Akaike Information Criterion adjusted for small sample % Pine forest % Extent of pinewoods % Eucalyptus % Extent ofEucalyptus woodland patches size) and considered as the final model the one with the lowest forest AICc value (Burnham and Anderson, 2002). All statistics were % Oakwood % Extent of patches of cork oak woodland implemented with the software R 2.9.2. (R Development Core forest Team, 2009) and all GLMs and LMMs were built throughglm the % Woodland % Extent of total woodland excluding pine, cork oak and and lme functions of the library (nlme). Before analysis, all propor- Eucalyptus forest % Greenhouses % Extent of strawberry greenhouses tions of land cover types were arcsine square root transformed to % Urban areas % Extent of urban areas conform to a normal distribution. All means are given ±1 SE, tests are two-tailed and statistical significance was pset6 0.05.at factor, individuals of all age and sex categories were tracked 3.simul- Results taneously (Table 1). We assume that such temporal overlap pre- vented biases associated with seasonal changes in habitat 3.1. Home range analysis selection. To reduce collinearity and the number of variables pre- sented to multivariate models, we employed the method of vari-The mean home range size for the pooled sample of individuals able reduction proposed by Green (1979) and commonly was 153.3 ± 28.6 2 (MCP;km range = 7.6–688.4 2,kmn = 38). For all employed in habitat selection studies (e.g. Sergio et al., 2003;methods Zuur of estimation, home range size varied among individuals Table 3 Mean estimates of home range size for male and female, breeding and floating Black kites in Doñana National Park (South-Western Spain). All ranges2 and are measured in km distances in metres.
Breeding males (range) Breeding females (range) Floatersa (range) Fb p Minimum Convex Polygon 80.0 ± 13.9 (23.4–164.5) 43.3 ± 11.6 (7.6–151.5) 310.2 ± 54.9 (104.8–688.4) 31.2 0.001 Kernel 50% 5.5 ± 1.0 (1.1–11.7) 2.7 ± 0.6 (0.7–9.2) 28.2 ± 7.4 (8.1–107.2) 35.1 0.001 Kernel 75% 13.4 ± 2.9 (2.3–33.6) 6.0 ± 1.6 (1.7–21.7) 83.0 ± 21.7 (20.1–326.0) 40.7 0.001 Kernel 95% 51.8 ± 11.7 (6.9–136.7) 17.9 ± 4.2 (5.0–58.0) 257.0 ± 58.6 (72.6–762.0) 39.8 0.001 Distance to centroid 2751.3 ± 332.5 (1209.0–4376.0) 1531.3 ± 225.1 (633.4–3278.5) 5738.8 ± 554.7 (3736.0–10715.5) 35.5 0.001
a Includes both males and females: the sample size was too small for testing sexual differences in floaters. b F-statistic from a one way ANOVA. of different sex and status: ranges were consistently largest for Table 5 floaters, smallest for breeding females and intermediate for breed-Generalised linear models (with binomial errors and a logit link function) discrim- ing males (Duncan’s post hoc ptest,< 0.05; Table 3). On average,inating between the 100 m buffers around the locations of radio-tracked Black kites floater ranges were 4–8 times larger than those of breeders. andSimi- an equal number of randomly generated locations. Models were built separately larly, the mean distance of all the radio-locations of an individualfor: (a) 12 breeding males, (b) 12 breeding females, (c) 14 non-breeding individuals (Doñana National Park, south-western Spain). In the table, n refers to the number of from the centroid of its range was largest for floaters, intermedradio-locations.iate for breeding males and shortest for breeding females (Duncan’s post hoc test,p < 0.05; Table 3). Parameter estimate ± SE t-Value p-Value (a) Male’s used vs. random locations (n = 1108) a 3.2. Larger scale habitat selection: whole home range level Marshland 0.22 ± 0.07 3.32 0.0009 Dehesaa 0.24 ± 0.07 3.60 0.0003 Scrublanda 0.22 ± 0.07 3.33 0.0009 Compared to random home ranges, those of breeding males Woodlandhad a 0.13 ± 0.07 1.95 0.0509 larger amounts of dehesa and marshland, and a lower incidenceFarmland of a 0.24 ± 0.07 3.52 0.0004 farmland, while those of breeding females had more scrubland Interceptand 0.69 ± 0.10 6.87 <0.001 less cultivation (Table 4). Finally, floater ranges had more marsh-(b) Female’s used vs. random locations (n = 1320) a land and scrubland than random ranges (Table 4). Marshland 0.24 ± 0.03 8.71 0.0405 Dehesaa 0.33 ± 0.03 10.59 0.0015 Scrublanda 0.31 ± 0.03 10.75 0.0038 3.3. Finer scale habitat selection: individual locations Farmlanda 0.15 ± 0.03 4.75 0.0041 Intercept 0.80 ± 0.04 20.94 <0.001 Similar results were obtained when focusing on the habitat(c) Floater’s used vs. random locations (n = 1532) composition in the 100 m buffers around each location (TableMarshland 5). a 0.30 ± 0.02 12.67 <0.001 a Compared to random locations, breeding males positively selectedDehesa 0.07 ± 0.03 2.13 0.0264 a the marshland, dehesa and scrubland, while avoiding woodlandScrubland 0.26 ± 0.03 9.29 <0.001 Farmlanda 0.13 ± 0.02 5.27 0.0475 and farmland (Fig. 2). Breeding females avoided cultivated fieldsIntercept 0.66 ± 0.03 19.97 <0.001 and positively selected the marshland, dehesa and scrubland, while floaters positively selected the marshland, dehesa and scrubland, a Proportion of each habitat in a 100 m buffer around each location (arcsin while avoiding farmland (Fig. 2). square-root transformed for analysis).
Table 4 Generalised linear models (with binomial errors and a logit link function) discrim- inating between the home ranges occupied by radio-tracked Black kites and an4. equalDiscussion number of randomly generated ranges. Models were built separately for: (a) 12 breeding males, (b) 12 breeding females, (c) 14 non-breeding individuals (DoñanaDuring the first few years of life, the floaters of solitary raptors, National Park, south-western Spain). such as many eagles, often live in so called temporary settlement Most competitive model Parameter estimate ± SE t-Value p-Value areas, which can be spatially separate and often far away from (a) Male’s occupied range vs. random range b(n = 24) breeding areas (e.g. Ferrer, 1993; Balbontín, 2005; Caro et al., Marshlanda 5.64 ± 3.67 29.32 0.047 2011). As a consequence, identifying and protecting suitable settle- Dehesaa 18.55 ± 9.85 22.29 0.008 ment areas requires major study efforts and can be extremely chal- a Farmland 18.02 ± 11.02 17.22 0.024 lenging (Penteriani et al., 2005). Contrary to this pattern, our study Intercept 9.73 ± 5.49 focused on a semi-gregarious species in which floaters, which were b (b) Female’s occupied range vs. random range (n = 24) mainly young individuals in their initial 1–7 years of life (Blas Marshlanda 1.18 ± 3.04 33.22 0.825* Dehesaa 2.69 ± 3.35 21.95 0.050 et al., 2009; Blas and Hiraldo, 2010), closely co-existed with terri- Scrublanda 6.57 ± 3.86 25.78 0.006 tory holders. It is difficult to say whether such radical difference Farmlanda 10.63 ± 6.04 17.17 0.029 among studies was exclusively caused by the social propensity of Intercept 4.52 ± 3.96 our species, but it would be extremely interesting to test whether (c) Floater’s occupied range vs. random range c(n = 28) the same pattern is found in other species of both groups (i.e. sol- a Marshland 18.32 ± 8.59 27.98 0.001 itarily breeding vs. colonial or loosely colonial species) and in other a Scrubland 21.90 ± 10.07 11.90 <0.001 bio-geographic regions. Independently of their cause, these results Intercept 18.32 ± 7.98 38.82 complete and extend our knowledge on the habitat and space a Proportion of each habitat in a home range (arcsin square-root transformedrequireme for nts of raptor floaters, up to now heavily biased towards analysis). solitary species with spatially separate floaters and breeders sec- b Includes the home ranges of 12 breeders and their 12 associated random ranges. tors of a population. c Includes the home ranges of 14 floaters and their 14 associated random ranges.Besides generalised local coexistence, the larger home ranges of * Included in the model with the lowest AICc value. floaters caused a marked ranging overlap with breeders (e.g. Fig. 3 6
Fig. 2.Mean percentage frequency of habitats observed around 1980 radio-locations of Black kites and around an equal number of randomly generated locations in Doñana National Park (South-Western Spain). and unpublished results). Such extremely large areas used byto kitemove than breeders (e.g. Carrete and Donazar, 2005; Guixé and floaters were possibly promoted by a combination of severalArroyo, fac- 2011). Second, floaters may roam over large areas to seek tors. First, floaters are not central place foragers and are thusterritory freer vacancies, test the breeders’ ability to defend their Fig. 3.Representative example of three home-ranges of Black kites of different sex and status in Doñana National Park (South-Western Spain) and its surroundings. The park location in Europe is portrayed in the inset. The solid black line represents the (smallest) home range of a breeding female, the black dotted line represents the (intermediate- sized) home range of a breeding male and the grey line with crosses represents the (largest) home range of a floater. The thin black line depicts the boundaries of the cumulated National and Natural Park (marine portion excluded). The seasonal marshland is shown in pale grey and intensive farmland in dark grey.
territories, and gain information on habitat quality for futuresize set- of a mosquito to a 1-kg adult rabbit, and use all sources of car- tlement (i.e., prospecting: Sergio and Penteriani, 2005; Whitfieldrion when available (authors’ personal observation). et al., 2009a,b; Sergio et al., 2011a). Third, floaters are on averageOn the other hand, despite their opportunistic diet, all kites younger and less experienced than breeders, they could thus forageclearly avoided agricultural habitats, mainly represented by rice less efficiently and need larger areas than breeders to gatherfields. the Such avoidance pattern was interesting given the kites’ gen- same daily amount of food. All these ideas are currently eralunder preference for aquatic habitats (Sergio et al., 2005) and given study through more radio-marking. that rice fields are inundated when the natural marshes start to With regard to breeding Black kites, the smaller ranges wedry re-in late spring-summer. Active avoidance by Black kites was corded were likely the result of central place foraging and thepossibly need related to the very intensive agricultural practices affect- for territory, mate and nest guarding. The home ranges of breedinging large areas around the park, which usually involve the use of females were about half the size of males, as expected in largeraptors amounts of broad-spectrum pesticides and herbicides well- where females are usually deputed to incubation, chick broodingknown to depress the populations of invertebrate species such as and nest guarding (e.g. Newton, 1979; Arroyo et al., 2009). crayfish, as well as reptiles, amphibians, mammals and birds (Law- Whatever the differences in the amplitude of home ranges,ler, it 2001;is Parsons et al., 2010), all of which are potential prey for interesting that individuals of all status categories basically kitesse- (e.g. Hiraldo et al., 1990; Viñuela and Veiga, 1997). The fact lected and avoided the same habitats. Black kites over-selectedthat intensive habitat transformations and farming deteriorate open semi-natural habitats (marshland, dehesa and scrubland),habitat quality and suitability for raptor populations has been and avoided close-structured or cultivated habitats (woodlandshown by numerous studies (e.g. Tella et al., 1998; Sánchez-Zapata and farmland). Such patterns were highly consistent across spatialand Calvo, 1999; Thiollay, 2006), although relationships can be scales of analysis, adding confidence to our results. The preferencomplexce and some species have also been shown to thrive well for open habitats can be explained by the aerial hunting strategiesin human-altered landscapes (e.g. Bird et al., 1996; Whitfield of the species and the fact that the three over-selected habitatset al., 2006b). hold important populations of the main prey species, such as waterbirds and crayfish for marshland and rabbits for mosaics4.1. ofImplications for conservation dehesa and scrubland (e.g. Hiraldo et al., 1990; Viñuela and Veiga, 1997). Besides their value in terms of prey abundance and distribu-The fact that kite breeders and floaters closely coexisted and tion, the homogeneous selection of these habitats across individu-showed similar habitat preferences opens interesting possibilities als may be promoted by the generalist and opportunistic feedingfor conservation strategies. Establishing habitat management habits of the species, which can prey upon live animals fromguidelines the that simultaneously favour individuals of different status would be relatively straightforward, and easier thanta inde Andalucía and by the Excellence Projects RNM 1790, RNM populations with separate breeding and settlement areas (e.g.3822 and RNM 7307 of the Junta de Andalucía. Ferrer and Harte, 1997; Balbontín, 2005). In the future, it would be interesting to test how frequently breeders and floaters Referencesshow similar habitat preferences in those cases in which they happen to coexist. On the other hand, the much larger ranges of Aebischer,floaters N.J., Robertson, P.A., Kenward, R.E., 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology 74, 1313–1325. implied that they used areas outside the national park moreArroyo, fre- B., Amar, A., Leckie, F., Buchanan, G., Wilson, J., Redpath, S., 2009. Hunting quently than breeders (e.g. Fig. 3), making them more susceptiblehabitat selection by hen harriers on moorland: implications for conservation to human-related disturbance, thus making their management management. Biol. Conserv. 142, 586–596. more challenging (e.g. through higher potential exposure to illegalBalbontín, J., 2005. Identifying suitable habitat for dispersal in Bonelli’s eagle: an important issue in halting its decline in Europe. Biol. Conserv. 126, 74–83. poisoned baits placed in private game reserves that surroundBird, the D., Varland, D., Negro, J.J., 1996. Raptors in Human Landscapes. Academic national park; Sergio et al., 2005; Tenan et al., 2012; authors’Press, London. unpublished results). Blanco, G., 1997. Role of refuse as food for migrant, floater and breeding Black Kites (Milvus migrans). J. Raptor Res. 31, 71–76. To date, the scanty information on habitat and range selectionBlanco, G., Lemus, J.A., Frías, O., Grande, J., Arroyo, B., Martínez, F., Beniandrés, M., by breeders and floaters of a raptor population has been mainly2007. fo- Contamination traps as trans-frontier management challenges: new cused on large, solitary species with disjoint settlement areasresearch on the impact of refuse dumps on the conservation of migratory scavengers. In: Cato, M.A. (Ed.), Environmental Research Trends. Nova Science which spatially separate the two sectors of the population. Our Publishers,re- New York, pp. 153–204. sults extend this notion to cases where breeders and floatersBlanco, spa- G., Viñuela, J., 2004. Milano Milvusnegro migrans. In: Madroño, A., González, tially coexist with similar habitat preferences. In both cases,C., Atienza, J.C. (Eds.), Libro Rojo de las Aves de España. MMA-SEO/BirdLife, Madrid, pp. 116–120. preserving the floater sector of a population, which may be aBlas, major J., Hiraldo, F., 2010. Proximate and ultimate factors explaining floating achievement to ensure long-term population persistence (Penteri- behavior in long-lived birds. Horm. Behav. 57, 169–176. ani et al., 2005), may present special challenges. On one hand,Blas, forJ., Sergio, F., Hiraldo, F., 2009. Age-related improvement in reproductive performance in a long-lived raptor: a cross-sectional and longitudinal study. solitary species with temporary settlement areas, identifying suchEcography 32, 647–657. sites may be difficult, labour intensive and duplicate the targetBlas, sur- J., Sergio, F., Wingfield, J.C., Hiraldo, F., 2011. Experimental tests of endocrine face in need of management. On the other hand, for social functionand in breeding and nonbreeding raptors. Physiol. Biochem. Zool. 84 (4), semi-social species, coexistence may ease the identification of local406–416. Boal, C.W., Andersen, D.E., Kennedy, P.L., 2005. Foraging and nesting habitat of targets, but the larger ranges of floaters may make them amenablebreeding male Northern Goshawks in the Laurentian mixed forest province, to different threats that may be subtle and difficult to identifyMinnesota. J. Wildl. Manag. 69, 1516–1527. without intensive study. For example, in a population of the Bosch,loosely R., Real, J., Tintó, A., Zozaya, E.L., Castell, C., 2010. Home ranges and patterns of spatial use in territorial Bonelli’s Eagles. Ibis 152, 105–117. colonial Red kiteMilvus ( milvus), floaters coexisted with breedersBurnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference: a but their wider ranging behaviour was proposed as the cause Practicalof Information-Theoretic Approach, second ed. Springer Press, New York. their higher propensity to die by poisoning and electrocutionCaro, J., Ontiveros, D., Pizarro, M., Pleguezuelos, J.M., 2011. Habitat features of settlement areas used by floaters of Bonelli’s and Golden Eagles. Bird Conserv. (Tavecchia et al., 2012). Similarly, in a population located closeInt. to 21, 59–71. a large rubbish dump, Black kite floaters coexisted with breedersCarrete, M., Donazar, J.A., 2005. Application of central-place foraging theory shows but with a diet more dominated by refuse, which may expose themthe importance of Mediterranean dehesas for the conservation of the cinereous vulture. Biol. Conserv. 126, 582–590. to toxic substances and pathogens (Blanco, 1997; Blanco etCastroviejo, al., J., 1993. Mapa del Parque Nacional de Doñana. Consejo Superior de 2007). Overall, independently of coexistence or spatial separation,Investigaciones Científicas y Agencia de Medio Ambiente de la Junta de the few data currently available suggest that the conservation Andalucía,of Madrid. Ciarniello, L.M., Boyce, M.S., Seip, D.R., Heard, D.C., 2007. Grizzly bear habitat the floater sector of raptor populations will continue to be chal-selection is scale dependent. Ecol. Appl. 17, 1424–1440. lenging and require special, ad hoc management plans. Finally,Clark, as T.W., Curlee, A.P., Minta, S.C., Kareiva, P.M., 1999. Carnivores in ecosystems: more studies will appear, the two alternatives of complete spatialthe Yellowstone experience. Yale University Press, New Haven, USA. separation with differential habitat selection and full coexistenceCrabtree, R.L., Sheldon, J.W., 1999. Coyotes and canid coexistence. In: Clark, T.W., Peyton Curlee, A., Minta, S.C., Kareiva, P.M. (Eds.), Carnivores in Ecosystems: the with shared habitat preferences are likely to emerge as the twoYellowstone ex- Experience. Yale University Press, New Haven, pp. 127–163. tremes of a more gradual continuum of breeders–floaters spatialDavid Smith, J.L., 1993. The role of dispersal in structuring the Chitwan tiger relations, each situation requiring locally fine-tuned conservation population. Behaviour 124, 165–195. Ellegren, H., 1996. First gene on the avian W chromosome (CHD) provides a tag for action. universal sexing of non-ratite birds. Proc. R. Soc. Lond. B 263, 1635–1641. As for the local Black kite population of Doñana NationalElkie, Park, P., Rempel, R., Carr, A., 1999. Patch Analyst User’s Manual: a Tool for the results of the foraging habitat analyses suggest a potentiallyQuantifying Landscape Structure. Ont. Min. Natur. Resour Northwest Science and Technol, Thunder Bay, Ontario. beneficial effect of the recently implemented restoration projectEstes, J.A., Terborgh, J., Brashares, J.R., Power, M.E., Berger, J., Bond, W.J., Carpenter, 2 ‘‘Doñana 2005’’, which converted some 50of kmfarmland (an S.R., Essington, T.E., Holt, R.D., Jackson, J.B.C., Marquis, R.J., Oksanen, T., Paine, avoided habitat) located inside the park to seasonal marshland R.T.,(a Pikitch, E.K., Ripple, W.J., Sandin, A.S., Scheffer, M., Schoener, T.W., Shurin, J.B., Sinclair, R.E., Soulé, M.E., Virtanen, R., Wardle, D.A., 2011. Tropic degrading preferred habitat) (García Novo and Marín Cabrera, 2005a). Furtherof planet earth. Science 333, 301–306. restoration of the vast expanses of marshland which were Fernández-Delgado,con- C., 2005. Conservation management of a European natural verted to agriculture through drainage in the 1960s, and whicharea: Doñana National Park, Spain. In: Groom, M.J., Meffe, G.K., Carroll, C.R. (Eds.), Principles of Conservation Biology. Sinauer Associated, Inc., currently surround the park, would be desirable and probably ben-Massachusetts, USA, pp. 536–543. eficial to this and other more exigent species. Ferrer, M., 1993. Juvenile dispersal behaviour and natal philopatry of a long-lived raptor, the Spanish Imperial EagleAquila adalberti. Ibis 135, 132–138. Acknowledgements Ferrer, M., Harte, M., 1997. Habitat selection by immature Spanish imperial eagles during the dispersal period. J. Appl. Ecol. 34, 1359–1364. García Novo, F., Marín Cabrera, C., 2005a. Doñana: Water and Biosphere. Doñana We thank F.J. Chicano, J.M. Giralt, and F.G. Vilches for help in2005 the Project. Spanish Ministry of the Environment, Madrid. field. We sincerely thank D. Oro, P. Whitfield and two anonymousGarcía Novo, F., Marín Cabrera, C., 2005b. The Doñana 2005 Project. In: García Novo, F., Marín Cabrera, C. (Eds.), Doñana: Water and Biosphere. Doñana 2005 Project. reviewers for their very detailed and illuminating comments thatSpanish Ministry of the Environment, Madrid, pp. 301–324. greatly improved a first draft of the manuscript. Part of thisGreen, study R., 1979. Sampling Design and Statistical Methods for Environmental was funded by Natural Research Ltd. and research projectsBiologists. Wiley Editions, West Sussex, England. Guixé, D., Arroyo, B., 2011. Appropriateness of Special Protection Areas for wide- CGL2008-01781 and CGL2011-28103 of the Ministerio de Cienciaranging species: the importance of scale and protecting foraging, not just e Innovación, JA-58 of the Consejería de Medio Ambiente de la nestingJun- habitats. Anim. Conserv. 14, 391–399. Hiraldo, F., Veiga, J.P., Mañez, M., 1990. Growth of nestling blackMilvus kites Sergio, F., Blas, J., Hiraldo, F., 2009. Predictors of floater status in a long-lived bird: migrans: effects of hatching order, weather and season. J. Zool. 222, 197–214.a cross-sectional and longitudinal test of hypotheses. J. Anim. Ecol. 78, 109– Hooge, P.N., Eichenlaub, B., 2000. Animal movement extension to Arcview. ver. 2.0.118. Alaska Science Center – Biological Science Office, U.S. Geological Survey,Sergio, F., Blas, J., López, L., Tanferna, A., Díaz-Delgado, R., Donázar, J.A., Hiraldo, F., Anchorage, AK, USA. 2011b. Coping with uncertainty: breeding adjustments to an unpredictable Kenward, R., 2001. A manual for Wildlife Radio Tagging. Academic Press, London.environment in an opportunistic raptor. Oecologia 166, 79–90. Lawler, S.P., 2001. Rice fields as temporary wetlands: a review. Israel J. Zool.Sergio, 47, F., Pedrini, P., Marchesi, L., 2003. Adaptive selection of foraging and nesting 513–528. habitat by black kitesMilvus ( migrans) and its implications for conservation: a Lehmann, E.L., Romano, J.P., 2005. Testing Statistical Hypotheses, third ed. Springer,multi-scale approach. Biol. Conserv. 112, 351–362. New York. Sergio, F., Penteriani, V., 2005. Public information and territory establishment in a Martín-López, B., García-Llorente, M., Palomo, I., Montes, C., 2011. The conservationloosely colonial raptor. Ecology 86, 340–346. against development paradigm in protected areas: valuation of ecosystemTavecchia, G., Adrover, J., Muñoz Navarro, A., Pradel, R., 2012. Modelling mortality services in the Doñana social–ecological system (southwestern Spain). Ecol. causes in longitudinal data in the presence of tag loss: application to raptor Econ. 70, 1481–1491. poisoning and electrocution. J. Appl. Ecol. 49, 297–305. Newton, I., 1979. Population Ecology of Raptors. T & AD Poyser, Berkhamsted,Tella, UK. J.L., Forero, M.G., Hiraldo, F., Donazar, J.A., 1998. Conflicts between Lesser Parsons, K.C., Mineau, P., Renfrew, R.B., 2010. Effect of pesticide use in rice fieldsKestrel on conservation and European Agricultural Policies as identified by habitat birds. Waterbirds 33, 193–218. use analyses. Conserv. Biol. 12, 593–604. Penteriani, V., Ferrer, M., Delgado, M.M., 2011. Floater strategies and dynamicsTenan, in S., Adrover, J., Muñoz Navarro, A., Sergio, F., Tavecchia, G., 2012. birds, and their importance in conservation biology: towards an understandingDemographic consequences of poison-related mortality in a threatened bird of nonbreeders in avian populations. Anim. Conserv. 14, 233–241. of prey. PLoS ONE 7 (11), e49187. http://dx.doi.org/10.1371/ Penteriani, V., Otalora, F., Sergio, F., Ferrer, M., 2005. Environmental stochasticity journal.pone.0049187.in dispersa areas can explain the ‘‘mysterious’’ disappearance of breeding Thompson, C.M., McGarigal, K., 2002. The influence of research scale on bald eagle populations. Proc. R. Soc. Lond. B 272, 1265–1269. habitat selection along the lower Hudson River, New York (USA). Landscape R Development Core Team, 2009. R: A language and environment for statisticalEcol. 17, 569–587. computing. Vienna, Austria.